I want to do a Human WGS analysis, below are my current server details. Can someone share from your experience that it would be enough or not.
Storage : 2 TB
OS : Ubuntu 20.4
Threads : 80
RAM : 256 GB
I want to do a Human WGS analysis, below are my current server details. Can someone share from your experience that it would be enough or not.
Storage : 2 TB
OS : Ubuntu 20.4
Threads : 80
RAM : 256 GB
If with "WGS" you mean mapping and variant calling then yes. Threads and RAM are sufficient (for a single sample). Question is how many samples you need to process in parallel in a given period of time. Storage can be a bottleneck at some point, but that is easy to upgrade. Even a 20TB server-grade HDD is affordable these days.
For one sample at a time that should be enough, although the storage looks very slim. I'd get an extra much bigger SSD and external backups in the cloud and or on external HDs (though these can fail, so the cloud is likely far better).
With 80 threads you can run more than one WGS at once, but RAM might be limiting for two on certain pipeline parts. A 2TB SSD will certainly be lacking but is cheap and easy to expand.
An ubuntu 20.04 system might be out of data by now given the standard 5 years of updates ran out in April, so upgrade if possible.
Thank you for the reply. I do have an other HDD of 4 TB. So, storage i can manage. My primary concern is RAM and threads, only due to heavy BAM processing.
As told, ubuntu is older version and due too TAT issue, I can't update the OS for now (Requires several other pipeline or package setting up later). So, for now I will try to manage with what i got.
Thanks again.
Theres heaps of IO so SSD speed is very important. You can monitor IO with a tool like bottom https://github.com/ClementTsang/bottom
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Thanks a lot for your suggestions. Yes, It is a for only sample. I do have a extra 4 TB HDD. So, storage i can manage i think. My only primary concern is RAM and threads.
Anyway, thanks again.
You also need to tell us how much data you have. If you had a billion (or two) reads then it may be a different thing but if you have standard 30-50 M then that is normal. Time it takes should be what you will need to look for.